Publication Bias in Systematic Reviews in International Development: Evidence from a Systematic Review on the Impact of Farmer Field Schools in Low- and Middle-Income Countries
Speaker: Hugh Waddington, Senior Evaluation Officer, 3ie
Date: January 24, 2013, 5-6 pm
Venue: Upper Meeting Room (103), LIDC, 36 Gordon Square, London
Systematic reviews of interventions aim to include the relevant literature addressing the research question regardless of publication status. If unpublished literature is not comprehensively searched, or is excluded from the review, different patterns of results between published and unpublished literature could bias the results of the review.
This presentation reports on the results of publication bias analysis for a systematic review on the impact of Farmer Field Schools interventions. The presentation demonstrates the importance of publication bias assessment in systematic reviews in international development. Both funnel plots and statistical test analysis provide evidence of publication bias, or small studies’ effect. Trim-and-fill analysis is conducted to quantify the extent of this bias and provide unbiased estimates of programme effects.
The results of the analysis highlight the importance of searching comprehensively the unpublished literature and suggest the advisability of conducting publication bias analysis to avoid the overestimation of programme effectiveness in systematic review results.